CN104603630B - Magnetic resonance imaging system with the motion detection based on omniselector - Google Patents

Abstract

The present invention provides a kind of magnetic resonance imaging system (200,300) for being used for acquisition of magnetic resonance data (242,244).For controlling processor (230) execute instruction (250,252,254,256,258) of the magnetic resonance imaging system, the instruction causes the processor repeatedly：(100) described magnetic resonance imaging system is controlled to gather the part of the MR data, wherein, each part of the MR data includes omniselector data (244)；The set of (102) omniselector vector is created by omniselector data described in each extracting section from the MR data；(104) Dissimilarity matrix (246,400,700,800,900,1000,1100,1400,1500) is built by the measurement between each in calculating the set of omniselector vector；Sort out (248) using subsumption algorithm to generate the matrix of (106) described Dissimilarity matrix；And (108) described magnetic resonance imaging system is controlled to sort out using the matrix to change the collection to the MR data.

Description

Magnetic resonance imaging system with navigator-based motion detection

technical field

The present invention relates to magnetic resonance imaging, and in particular to the modification of the acquisition of magnetic resonance imaging using dissimilarity matrix sorting.

Background technique

Patient motion during magnetic resonance imaging (MRI) data acquisition can cause image artifacts that compromise the diagnostic quality of the resulting images. This is an important issue and has given rise to a number of approaches aimed at reducing the effects of patient motion. Despite these past efforts, patient motion remains an important issue today; in part due to the increased SNR and fast scanning methods in modern magnetic resonance (MR) scanners that allow imaging at higher spatial resolution, making The experiments are more sensitive to motion, partly because the proposed method is impractical for routine clinical application. This may be the case because they require sensors that are too complex and/or time-consuming to connect, or because they unduly extend the scan duration.

Other common limitations are restrictions on certain anatomies/imaging sequences, negative effects on image contrast, etc.

The method of cluster analysis is explained in "The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second Edition (SpringerSeries in Statistics)" by Trevor Hastie, Robert Tibshirani, Jerome Friedman, Chapter 14.3.

Contents of the invention

The invention provides a magnetic resonance imaging system and a computer program product in the independent claims. Embodiments are given in the dependent claims.

As will be appreciated by those skilled in the art, aspects of the invention may be implemented as an apparatus, method or computer program product. Accordingly, aspects of the present invention can take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, resident software, microcode, etc.) referred to as "circuits", "modules" or "systems". Furthermore, aspects of the invention may take the form of a computer program product embodied in one or more computer-readable media having computer-executable code embodied thereon.

A combination of one or more computer readable medium(s) may be utilized. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. As used herein, 'computer-readable storage medium' encompasses any tangible storage medium that can store instructions executable by a processor of a computing device. The computer-readable storage medium may be referred to as a computer-readable non-transitory storage medium. The computer readable storage medium may also be referred to as a tangible computer readable medium. In some embodiments, a computer-readable storage medium can also store data that can be accessed by a processor of a computing device. Examples of computer readable storage media include, but are not limited to: floppy disks, magnetic hard drives, solid state drives, flash memory, USB thumb drives, random access memory (RAM), read only memory (ROM), optical disks, magneto-optical disks, and processor storage file. Examples of optical disks include compact disks (CDs) and digital versatile disks (DVDs), such as CD-ROM, CD-RW, CD-R, DVD-ROM, DVD-RW or DVD-R disks. The term computer-readable storage medium also refers to various types of recording media that can be accessed by computer devices via a network or communication link. For example, data can be retrieved on a modem, on the Internet, or on a local area network. Computer-executable code embodied on a computer readable medium may be transmitted using any suitable medium, including but not limited to wireless, telephone line, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

A computer readable signal medium may include a propagated data signal with computer executable code embodied therein, eg, in baseband or as part of a carrier wave. Such a propagated signal may take any of various forms, including, but not limited to, electromagnetic, optical, or any suitable combination thereof. A computer-readable signal medium may be any computer-readable medium, which is not a computer-readable storage medium, and which is capable of communicating, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device .

'Computer storage' or 'memory' are examples of computer readable storage media. Computer memory is any memory that is directly accessible to the processor. 'Computer storage' or 'storage' are further examples of computer-readable storage media. A computer storage device is any non-volatile computer-readable storage medium. In some embodiments, computer storage may also be computer memory, and vice versa.

'Processor' as used herein encompasses an electronic component capable of executing a program or machine-executable instructions or computer-executable code. References to a computing device including a "processor" should be interpreted as possibly including more than one processor or processing core. For example, the processor may be a multi-core processor. A processor may also refer to a collection of processors within a single computer system or distributed among multiple computer systems. The term computing device should also be interpreted as possibly referring to a collection or network of computing devices, each computing device including one or more processors. The computer-executable code may be executed by multiple processors, which may be within the same computing device, or may even be distributed across multiple computing devices.

Computer-executable code may comprise machine-executable instructions or programs that cause a processor to perform an aspect of the invention. Computer-executable code for carrying out operations for various aspects of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, Smalltalk, C++, etc., as well as conventional procedural programming languages language, such as the "C" programming language or a similar programming language, and are compiled into machine-executable instructions. In some instances, the computer-executable code may be in a high-level language or in precompiled form, and used in conjunction with an interpreter that generates the machine-executable instructions online.

The computer-executable code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the user's computer. Execute on a remote computer or server. In the latter case, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or a connection may be established to an external computer (for example, by using an Internet service provider's Internet).

Aspects of the present invention are described with reference to flowchart illustrations, illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It should be understood that each block or sub-blocks in the flowcharts, illustrations and/or block diagrams can be implemented by computer program instructions, in the form of computer executable code where applicable. It is also to be understood that combinations of blocks from different flowcharts, illustrations and/or block diagrams may be combined where not mutually exclusive. These computer program instructions may be provided to a processor of a general purpose computer, a special purpose computer, or other programmable data processing apparatus to produce a machine whereby said The instructions create modules for implementing the functions/actions specified in the flowchart illustrations and/or block diagram block(s).

These computer program instructions may also be stored on a computer-readable medium capable of instructing a computer, other programmable data processing apparatus, or other equipment to function in a specific manner such that the instructions stored on the computer-readable medium An article of manufacture is produced that includes instructions for implementing the functions/acts specified in the flowchart and/or block(s) of the block diagram.

The computer program instructions can also be loaded into a computer, other programmable data processing device or other equipment, so as to realize a series of operation steps executed on the computer, other programmable device or other equipment, so as to generate a computer-implemented process such that the instructions executed on the computer or other programmable apparatus provide a process for implementing the functions/actions specified in the flowchart and/or block(s) in the block diagram.

A 'user interface' as used herein is an interface that allows a user or operator to interact with a computer or computer system. A 'user interface' may also be referred to as a 'human-computer interaction device'. A user interface may provide information or data to and/or receive information or data from an operator. The user interface may enable input from an operator to be received by the computer and may provide output from the computer to the user. In other words, the user interface may allow an operator to control or manipulate the computer, and the interface may allow the computer to indicate the effect of the operator's control or manipulation. The display of data or information on a display or graphical user interface is an example of providing information to an operator. From keyboards, mice, trackballs, touchpads, pointing sticks, graphics tablets, joysticks, gamepads, webcams, headsets, gear levers, steering wheels, pedals, wired gloves, dance mats, remote controls, and accelerometers are all examples of self-control An example of a user interface component for the receipt of information or data from the recipient.

As used herein, 'hardware interface' encompasses an interface that enables a processor of a computer system to interact with, and/or control, external computing devices and/or devices. A hardware interface may allow the processor to send control signals or instructions to external computing devices and/or devices. Hardware interfaces may also enable the processor to exchange data with external computing devices and/or devices. Examples of hardware interfaces include, but are not limited to: Universal Serial Bus, IEEE 1394 port, parallel port, IEEE 1284 port, serial port, RS-232 port, IEEE-488 port, Bluetooth connection, wireless LAN connection, TCP/IP connection, Ethernet connection, control voltage interface, MIDI interface, analog input interface and digital input interface.

As used herein 'display' or 'display device' encompasses an output device or user interface suitable for displaying images or data. The display can output visual, audio and or tactile data. Examples of displays include, but are not limited to: computer monitors, television screens, touch screens, tactile electronic displays, braille screens, cathode ray tubes (CRTs), memory tubes, bi-stable displays, electronic paper, vector displays, flat panel displays, vacuum fluorescent Display (VF), Light Emitting Diode (LED) Display, Electro Luminescent Display (ELD), Plasma Display Panel (PDP), Liquid Crystal Display (LCD), Organic Light Emitting Diode Display (OLED), Projector, and Head Mounted Display.

Magnetic resonance (MR) data is defined herein as the recorded measurements of radio frequency signals emitted by the antenna of a magnetic resonance apparatus via atomic spins during a magnetic resonance imaging scan. Magnetic resonance data is an example of medical image data. A magnetic resonance imaging (MRI) image is defined herein as a reconstructed two-dimensional or three-dimensional visualization of anatomical data contained within said magnetic resonance imaging data. The visualization can be performed using a computer. Portions of magnetic resonance data may also be referred to as "shots". Navigator data is an example of magnetic resonance data and typically represents the position or motion state of an object.

In an aspect of the invention, a magnetic resonance imaging system for acquiring magnetic resonance data from an imaging region is provided. The magnetic resonance imaging system includes a processor for controlling the magnetic resonance imaging system. The magnetic resonance imaging system also includes a memory for storing machine executable instructions for execution by the processor. Execution of the machine-executable instructions causes the processor to repeatedly control the magnetic resonance imaging system to acquire portions of the magnetic resonance data. Each portion of the magnetic resonance data includes navigator data. For some magnetic resonance imaging protocols, the data may be acquired over a period of several minutes. Said portion of said magnetic resonance data refers to the portion of magnetic resonance data acquired during a complete protocol.

Navigator data as used herein encompasses magnetic resonance data indicative of motion of a subject. For example, if the object is completely stationary inside and outside, the navigator data should not change. However, if the object is moving or is moving internally, the navigator data can be used to represent or quantify this motion. In some embodiments, the navigator data may be separately acquired navigator data acquired in an interleaved manner with the portion(s) of the magnetic resonance data. The navigator data can also be image data and/or data in k-space which are extracted from the part of the magnetic resonance data. Different types of navigators are available. For example, the navigator data may be one-dimensional signals or projections, such as central k-space lines. F navigator (k-space line parallel to central k-space line), O navigator (circle around k-space origin) are examples. Measurements of higher dimensional data (such as small images or subsets of them) can also be used as navigators.

Execution of the machine-executable instructions further causes the processor to repeatedly create a set of navigator vectors by extracting the navigator data from each portion of the magnetic resonance data. After each acquisition of said navigator data, a navigator vector is constructed and then added to the set of navigator vectors.

Execution of the machine-executable instructions further causes the processor to repeatedly build a dissimilarity matrix or D-matrix by computing a metric between each of the set of navigator vectors. A metric as used herein is a mathematical function used to generate a value(s) to evaluate the proximity or similarity of two navigator vectors. As used herein, a dissimilarity matrix encompasses a matrix for storing relative dissimilarity as measured by a metric between two or more mathematical objects (eg, navigator vectors).

The dissimilarity between each of the navigator vectors is measured using the metric, and this is then used to construct the dissimilarity matrix. As more and more navigator data is collected, the dissimilarity matrix may not need to be fully reconstructed each time. For example, when adding a navigator vector, additional data can simply be added to the existing dissimilarity matrix, making it larger. Execution of the instructions further causes the processor to repeatedly generate matrix classifications of the dissimilarity matrices using a classification algorithm. The classification algorithm takes as input the dissimilarity matrix and then outputs a matrix classification. The classification algorithm can thus return a classification describing the structure of the values inside the dissimilarity matrix, or can identify transitions in the values of the navigator set. Execution of the instructions further causes the processor to control the magnetic resonance imaging system to modify the acquisition of the magnetic resonance data using the classification matrix. For example, the matrix classification may indicate that the object has moved internally or externally since the last data acquisition. This can be used, for example, to determine whether the magnetic resonance protocol should be continued or whether it should be interrupted. This embodiment may be beneficial as it enables the determination of various types of movement inside or outside the object. This can help generate magnetic resonance images more quickly.

In another embodiment, said classification algorithm is a pattern recognition algorithm that can be used to select said classification matrix. For example, the pattern recognition algorithm may be programmed or contain instructions that enable the process to recognize transitions in the values of the navigator vectors, or may be operable to recognize patterns in a collection of navigator vectors. For example, a neural network can be programmed to classify the matrix.

In another embodiment, the memory further includes a library of matrices including example matrices. The pattern recognition module can be used to select one of the example matrices. This embodiment can be beneficial because it enables selection of a specific library matrix. This can help generate the matrix classification more quickly.

In another embodiment, each of said example matrices is associated with a modification instruction. Acquisition of the magnetic resonance data by the magnetic resonance imaging system is modified by executing the modification instructions. This embodiment may be beneficial because it provides a means of modifying the acquisition of the magnetic resonance data on-line while the acquisition is taking place. The use of matrices in this context can be useful because the examples in the matrix library can be well formulated and the modifications to improve data quality can be known.

In another embodiment, the pattern recognition algorithm is a cluster analysis algorithm. The cluster analysis algorithm can be used to perform a temporal correlation of the set of navigator vectors. The temporal correlation between the various navigator vectors can then be used to generate the dissimilarity matrix. Cluster analysis algorithms as used herein encompasses algorithms that can be used for cluster analysis (which is also known as data partitioning). This can be particularly beneficial, as cluster analysis can be useful to classify the navigator vectors into different groups. For example, if the object has moved, the cluster analysis algorithm may identify that there are two or more distinct groups of navigator vectors. By dividing the navigator vectors into such groups it is then easy to identify when an object is moving. However, in some instances, the cluster analysis may indicate that the navigator vectors have a higher degree of similarity to a certain period. For example, this may indicate a periodic motion of the object.

In another embodiment, the clustering algorithm is an agglomerative hierarchical clustering algorithm using average linkage. This is an iterative process, where you start with a cluster containing only one element (=one navigator fire). This means that at the beginning, the clustered data is equal to the number of navigator fires. In each iteration, the number of clusters is reduced by one by merging the two clusters that are closest to each other.

In order to use the described method, you need to define a metric that calculates the distance between two clusters. There are also different options, but all based on the dissimilarity matrix values for the elements belonging to said clusters: you can take the minimum, maximum or average of the dissimilarity values of all elements involved.

This gives you a sequence of gradually decreasing clusters with progressively increasing inter-cluster dissimilarity (where fusion between two clusters occurs). If you plot fused dissimilarity against the number of clusters, you will notice strong jumps in the plot if navigator firings can be separated into distinct groups. By placing the threshold at the maximum allowable inter-cluster distance, you select a specific cluster from the hierarchical chain of clusters.

In another embodiment, the classification algorithm is a statistical analysis algorithm. This can be beneficial because there are a variety of well-known statistical techniques that can be used to identify changes in data.

In another embodiment, the statistical analysis algorithm can be used to determine matrix classification by performing any of the following: performing Bayesian analysis, thresholding the dissimilarity matrix, computing the phase standard deviation of the dissimilarity matrix, identifying elements in the dissimilarity matrix that are outside a predetermined range, and performing probability-based selection.

In another embodiment, the magnetic resonance data comprises a plurality of slices. Execution of the instructions further causes the processor to compute, for each of the plurality of slices, a set of dissimilarity matrices using the dissimilarity matrix. Execution of the instructions further causes the processor to generate a set of matrix classifications by using the classification algorithm to generate the dissimilarity matrix for each of the set of dissimilarity matrices. Execution of the instructions further causes the processor to control the magnetic resonance imaging system to modify the acquisition of the magnetic resonance data using a set of matrix classifications. When acquiring magnetic resonance data, it may be acquired in more than one acquisition plane or slice. A slice as used herein covers the two-dimensional region from which magnetic resonance data is acquired. References to 2D are in fact references to desired positions. The magnetic resonance data is acquired in Fourier space, thus the acquisition of magnetic resonance data is essentially from a three-dimensional volume. According to each specific slice, its own dissimilarity matrix can be used. This embodiment may be particularly beneficial as it allows optimization of the magnetic resonance protocol in each specific slice.

In another embodiment, said controlling the magnetic resonance imaging system is performed by applying a rule-based algorithm to the set of matrix classifications.

In another embodiment, the magnetic resonance data comprises a set of sample points in k-space. Each portion of the magnetic resonance data is a subset of a set of sample points.

In another embodiment, the magnetic resonance imaging system further comprises a multi-channel radio frequency system operable to receive said magnetic resonance data from more than one channel simultaneously. Execution of the instructions further causes the processor to create a set of navigator vectors by combining the navigator data from the more than one channel. For example, navigator data can be collected on each individual data. The navigator data from each of the channels can be combined using a number of different techniques. For example, typically in multi-channel radio frequency systems, the data from each of the channels is weighted differently for different geographic locations. This weighting factor between opposing channels can be used to combine sets of navigator vectors. In other embodiments, the navigator vectors may simply be averaged. In other embodiments, a subset of said navigator data collected from said different radio frequency channels may be selected.

In another embodiment, said navigator data is combined by using any of the following: averaging said navigator data from said more than one channel using predetermined weights, and concatenating data from said navigator data of said more than one channel. In another embodiment, the predetermined weighting factor is any one of: Roemer sensitivity, spatial location, or received signal strength.

Concatenation enables the reduction of a large number of measurements to a single dissimilarity value. One way to achieve this is to define a metric ("d") for each measurement ("x _ij "), and then sum the values of all metrics applied to their respective measurements. In this case, the i subscripts may refer to different navigator excitations, and the j subscripts may refer to different measurement values. These can be different k-space samples of the same reception channel or different reception channels or even different devices (eg breathing belts).

In another embodiment, in the case of multi-channel magnetic resonance acquisitions, the navigator signals can be combined to a further process using appropriate channel communication, this approach implying certain weighting factors for reducing the value during evaluation workload. These weighting factors can be derived from potentially available coil sensitivity information and also from a priori knowledge to additionally balance the influence of the individual signals of the navigator data. This can be in terms of space and confidence. In another embodiment, this may include omitting the signal combination process by specifying a suitable metric that takes into account the potentially multi-channel nature of the magnetic resonance navigator signals during calculation of the matrix elements .

In another embodiment, said acquisition of magnetic resonance data is modified by any of: stopping said acquisition of magnetic resonance data; modifying scan geometry and limiting said acquisition of magnetic resonance data ; ignoring one or more points in the magnetic resonance data; reacquiring portions of the magnetic resonance data; generating an operator alert; and combinations thereof. Various actions may be required for correcting the acquisition of the magnetic resonance data, as determined by the matrix classification.

In another embodiment, the metric is any of the following: calculating the sum of squared compound differences between navigator vectors; calculating the difference in magnitude of said navigator vectors; calculating the difference between the navigator vectors and, computing the correlation between the navigator signals. The measuring may also include normalizing the navigator vectors. For example, the navigator vectors may be normalized using the minimum value of the D-matrix as part of calculating the metric.

When performing metric calculations on two navigator vectors, in some embodiments a Fourier transform may be performed on the navigator data as a pre-processing step.

The navigator data may be one-dimensional signals or projections, such as central k-space lines. F-navigators (k-space lines parallel to the central k-space line), O-navigators (circles around the k-space origin) are examples. Measurements of data in higher dimensions (such as small images or subsets thereof) can also be used as navigators.

In another embodiment, the magnetic resonance data also includes image data. Execution of the instructions causes the magnetic resonance imaging system to acquire the image data from a first region of interest and acquire the navigator data from a second region of interest. In an alternative embodiment, said navigator data is derived from said image data. In some embodiments, the navigator may be magnetic resonance signals derived from the actual imaging and/or volume, or may alternatively be derived from a sub-volume different from the imaging volume.

In another embodiment, the magnetic resonance imaging system comprises a motion detection system for generating motion data/navigator type data. Execution of the instructions causes the process to acquire the motion data during the acquisition of the magnetic resonance data. Execution of the instructions further causes the processor to incorporate the motion data into the dissimilarity matrix. This embodiment can be beneficial because additional data can be used to compose the dissimilarity matrix, making it more accurate. For example, the signal may result from an external motion sensing device, such as a pillow, belt or camera placed outside the patient. In some embodiments, the sources of the navigators may refer to the same motion state, and the sources of the navigators are taken into account simultaneously using a suitable dissimilarity matrix supporting this.

To combine different types of data into a single dissimilarity matrix, a scaling factor or other functions can be used.

Combining navigator data from multiple sources or using motion monitoring systems such as breathing belts, data is not the fundamental issue. In practice, it would of course require some experimentation to balance the influence of the different measurements relative to each other (ie to adjust the metric).

The motivation for combining navigator data from different elements is slightly different from joins. The work for the calculation of the dissimilarity matrix scales the quadratic equation with the number of navigator firings. For fast calculation of the D-matrix values, it is advantageous to keep the navigator vectors short. This is why it can be useful to combine the navigator vectors into one vector before computing the dissimilarity. Even greater compression can be applied: for example, the ability to average adjacent samples, or only take every nth sample into account.

In another aspect of the invention there is provided a computer program product comprising machine readable instructions for execution by a processor controlling a magnetic resonance imaging system for acquiring magnetic resonance data from an imaging region. Execution of the machine-executable instructions causes the processor to repeatedly control the magnetic resonance imaging system to acquire portions of the magnetic resonance data. Each portion of the magnetic resonance data includes navigator data.

Execution of the machine-executable instructions further causes the processor to repeatedly create a set of navigator vectors by extracting the navigator data from each portion of the magnetic resonance data. Execution of the machine-executable instructions further causes the processor to repeatedly generate matrix classifications of the dissimilarity matrices using a classification algorithm. Execution of the machine-executable instructions further causes the processor to repeatedly control the magnetic resonance imaging system to modify the acquisition of the magnetic resonance data using the matrix sorting.

In another aspect of the present invention, a method for controlling the magnetic resonance imaging system is also provided. The method comprises the step of repeatedly controlling the magnetic resonance imaging system to acquire portions of the magnetic resonance data. Each portion of the magnetic resonance data includes navigator data. The method further comprises the step of creating a set of navigator vectors by extracting the navigator data from each portion of the magnetic resonance data. The method also includes the step of constructing a dissimilarity matrix by computing a measure between each of the set of navigator vectors. The method further comprises the step of generating a matrix classification of said dissimilarity matrix using a classification algorithm. The method further comprises the step of controlling the magnetic resonance imaging system to modify the acquisition of the magnetic resonance data using the matrix classification.

It will be appreciated that one or more of the foregoing embodiments of the invention may be combined as long as the combined embodiments are not mutually exclusive.

Description of drawings

Preferred embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings, in which:

Figure 1 shows a flowchart illustrating a method according to an embodiment of the invention;

Fig. 2 illustrates an example of a magnetic resonance imaging system according to an embodiment of the invention.

Fig. 3 shows a magnetic resonance imaging system according to other embodiments of the present invention;

Figure 4 shows a visualization of the dissimilarity matrix for an example from T2-weighted head imaging;

FIG. 5 shows another visualization of a dissimilarity matrix 500 also from T2-weighted head imaging;

Figure 6 shows another visualization of the dissimilarity matrix constructed during T1-weighted shoulder imaging, and also illustrates periodic motion;

Figure 7 shows another example of a dissimilarity matrix acquired during T1-weighted shoulder imaging;

Figure 8 shows another visualization of the dissimilarity matrix illustrating sudden irreversible motion during T1-weighted shoulder imaging;

Figure 9 shows another visualization of the dissimilarity matrix illustrating how image quality can be improved by analyzing the dissimilarity matrix;

Figure 10 shows another visualization of the dissimilarity matrix;

Figure 11 shows a visualization of the matrix of Figure 10 after reordering;

Figure 12 shows a graph of the number of clusters 1200 versus fusion threshold 1202 for the examples shown in Figures 10 and 11;

Figure 13 depicts the distribution of cluster sizes depending on the level of dissimilarity for the examples shown in Figures 10 and 11;

Figure 14 shows another visualization of the dissimilarity matrix;

Figure 15 shows another visualization of the dissimilarity matrix;

Figure 16 shows the clustering index for different TSE excitations for the dissimilarity matrix shown in Figure 14;

Figure 17 shows the clustering index for different TSE excitations for the dissimilarity matrix shown in Figure 15;

Figure 18 shows the evolution of the transition probability distribution (PDF) during the experiment for the examples shown in Figures 14 and 15; and

FIG. 19 illustrates the evolution of knowledge about transition probabilities between two observed motion states in the two examples shown in FIG. 18 .

List of reference signs:

200 Magnetic Resonance Imaging Systems

204 magnets

206 bore diameter of magnet

208 imaging area

210 magnetic field gradient coil

212 Magnetic Field Gradient Coil Power Supply

214 RF coil

216 transceivers

218 objects

220 object supports

226 Computer systems

228 hardware interface

230 processor

232 user interface

236 Computer Storage Devices

238 computer memory

240 pulse train

242 Magnetic resonance data

244 navigator data

246 Dissimilarity Matrix

248 Matrix classification

250 control module

252 navigator data extraction module

254 dissimilarity matrix generation module

256 matrix classification module

300 Magnetic Resonance Imaging System

302 motion detection system

304 First Region of Interest

306 Second ROI

310 matrix library

312 Modification order

314 sports data

400 dissimilarity matrix

402 number of excitations

404 number of triggers

406 excite 5

700 dissimilarity matrix

702 excite 95

800 dissimilarity matrix

802 inspire130

804 excites 170 to 180

900 dissimilarity matrix

902 Inspire 1 to 4

1000 dissimilarity matrix

1100 dissimilarity matrix

1200 Number of clusters

1202 fusion threshold

1300 Dissimilarity

1302 fusion threshold

1400 dissimilarity matrix

1500 dissimilarity matrix

1800 left image

1802 right image

1900 left image

1902 right image

detailed description

Like numbered elements in these figures are equivalent elements or perform the same function. Elements already discussed previously are discussed in subsequent figures if they are functionally equivalent.

Fig. 1 shows a flowchart illustrating a method according to an embodiment of the invention. Magnetic resonance data are first acquired in step 100 . The magnetic resonance system acquires a portion of the magnetic resonance data. Each portion of the magnetic resonance data includes navigator data. Next, in step 102, a set of navigator vectors is created by extracting navigator data from each portion of the magnetic resonance data. Next, in step 104, a dissimilarity matrix is calculated using the measure between each of the set of navigator vectors. Since this is an iterative process, the dissimilarity matrix can be updated from the previous version simply by adding newly acquired navigator vectors and computing the dissimilarity between existing navigator vectors. Next, in step 106, a matrix classification of the dissimilarity matrix is generated using a classification algorithm. Then, in step 108, said acquisition of said magnetic resonance data is modified using said matrix classification. For example, the matrix classification may be associated with a set of commands for modifying how the magnetic resonance data is acquired. The method then proceeds to step 100, and the steps are repeated until all of the magnetic resonance data has been acquired.

Fig. 2 illustrates an example of a magnetic resonance imaging system 200 according to an embodiment of the invention. The magnetic resonance imaging system 200 includes a magnet 204 . The magnet 200 is a superconducting cylindrical magnet 200 having an aperture 206 therethrough. The use of different types of magnets is also possible. For example, it is also possible to use both split cylindrical magnets and so-called open magnets. A split cylindrical magnet is similar to a standard cylindrical magnet except that the cryostat has been split in two to allow access to the isoplane of the magnet, such magnets may be used in conjunction with charged particle beam therapy, for example. An open magnet has two magnet parts, one above the other, with a space between them large enough to accommodate the object: the arrangement of the two part areas is similar to that of a Helmholtz coil. Open magnets are favored because the subject is less constrained. Inside the cryostat of the cylindrical magnet is a set of superconducting coils. Within the bore 206 of the cylindrical magnet 204 there is an imaging zone 208 in which the magnetic field is strong and uniform enough to perform magnetic resonance imaging.

Also within the aperture 206 of the magnet is a set of magnetic field gradient coils 210 which are used for the acquisition of magnetic resonance data to spatially encode magnetic spins within the imaging region 208 of the magnet 204 . The magnetic field gradient coils 210 are connected to a magnetic field gradient coil power supply 212 . Magnetic field gradient coils 210 are considered representative. Typically, magnetic field gradient coils 210 comprise three separate sets of coils for spatially encoding in three orthogonal spatial directions. A magnetic field gradient power supply applies current to the magnetic field gradient coils. The current applied to the magnetic field gradient coils 210 is controlled as a function of time and may be ramped or pulsed.

Adjacent to the imaging region 208 is a radio frequency coil 214 for manipulating the orientation of the magnetic spins within the imaging region 208 and for receiving radio frequency transmissions from the spins also within the imaging region 208 . RF antennas may contain multiple coil elements. The radio frequency antenna may also be referred to as a channel or antenna. The radio frequency coil 214 is connected to a radio frequency transceiver 216 . The radio frequency coil 214 and radio frequency transceiver 216 may be replaced by separate transmit and receive coils and separate transmitter and receiver. It should be understood that RF coil 214 and RF transceiver 216 are representative. The radio frequency coil 214 is considered to represent also a dedicated transmit antenna and a dedicated receive antenna. Similarly, transceiver 216 may also represent a separate transmitter and receiver. The RF coil 214 may also have multiple receive/transmit elements, and the RF transceiver 216 may have multiple receive/transmit channels.

Magnetic field gradient coil power supply 212 and transceiver 216 are connected to hardware interface 228 of computer system 226 . The computer system 226 also includes a processor 230 . Processor 230 is connected to hardware interface 228 , user interface 232 , computer storage 234 and computer memory 236 .

Computer memory 236 is shown containing control module 250 . The control module 250 contains computer executable code that enables the processor 230 to control the operation and functions of the magnetic resonance imaging system 200 . For example, control module 250 may use pulse sequence 240 to acquire magnetic resonance data 242 . Computer memory 236 is also shown as containing navigator data extraction module 252 . Navigator data extraction module 252 contains computer executable code that enables the processor to extract navigator data 244 from magnetic resonance data 242 . The exact implementation of module 252 may depend on the nature of navigator data 244 . For example, module 252 may extract portions of k-space data from magnetic resonance data 242 . Computer memory 236 is shown as also containing dissimilarity matrix generation module 254 . Module 254 contains computer executable code that enables processor 230 to generate dissimilarity matrix 246 using navigator data 244 . Computer memory 236 is also shown as containing matrix classification module 256 . Matrix classification module 256 includes computer executable code that enables processor 230 to generate matrix classification 248 using dissimilarity matrix 246 . Computer memory 236 is also shown as containing pulse sequence modification module 258 . Pulse sequence modification module 258 contains computer executable code that enables processor 230 to modify pulse sequence 240 . Modifications to the pulse sequence may or may not imply modification of the individual commands executed as specified by the pulse sequence 240 . For example, the pulse sequence modification module 258 may simply have the portion of the data being acquired.

FIG. 3 shows a magnetic resonance imaging system 300 that is similar to the magnetic resonance imaging system shown in FIG. 2 . In this embodiment, there is an additional motion detection system 302 . Motion detection system 302 is considered to represent any system capable of detecting motion of object 218 . 302 may be, for example, a camera, sensor, accelerometer, or other type of motion detection system. Illustrated within the imaging region 208 is a first region of interest 304 from which magnetic resonance data 242 is acquired. Also shown within the imaging region 208 is a second region of interest 306 . Navigator data 244 is collected from second region of interest 306 . In some embodiments, the second region of interest 306 is at least partially within the first region of interest 304 . It should be noted that the magnetic resonance data navigator data is acquired in Fourier space. As such, the magnetic resonance data and the Fourier data are acquired at least partially outside the first region of interest 304 and the second region of interest 306 .

Computer storage 234 is also shown as containing matrix library 310 . The library of matrices contains a selection of example matrices that can be compared with the dissimilarity matrix 246 by the matrix classification module 256 . A matrix from library 310 is selected as the best matching dissimilarity matrix 246 . In this embodiment, there is a modification instruction 312 associated with the matrix classification 248 . The modification instructions 312 are then used by the control module 250 to modify the pulse sequence 240 . Computer storage 234 is also shown as containing motion data 314 collected using motion detection system 302 . The dissimilarity matrix generation module 254 is operable in this embodiment to cause the processor to incorporate the motion data 314 into the dissimilarity matrix 246 .

Embodiments of the present invention may provide a general method for real-time characterization of patient motion based on navigator signals. It can especially solve the following problems:

1. It enables differentiation between a number of different motion types. For example, between periodic movements such as regular breathing movements, sudden occasional patient movements such as coughing, and irreversible movements that substantially change the patient's position. Being able to make this distinction is important because it can be used to predict the success or failure of continued data acquisition and can help trigger appropriate decisions on how to proceed with scanning.

2. It can analyze motion during the actual data acquisition session and does not require calibration. In particular, it does not rely on assumptions made in other motion correction techniques, such as:

- Movement during the scan is the same as during the Navigator "training phase" preceding said scan

- There is an "average state" that can be used as a reference in motion analysis

- Motion below a certain threshold is permissible, while above the threshold is not.

Experience has shown that there are limits to reliance on any of these assumptions.

3. It can be used to predict how useful an appropriate reacquisition of the data is to reduce/avoid motion artifacts.

4. The method is fast. It can be performed during measurements and can be used to control data acquisition in real-time, helping to decide which part of the data should be re-acquired to further improve data consistency and image quality.

5. No planning for the navigator is required and it can be performed with numerous motion sensing approaches.

Embodiments of the invention may have the following features:

1. Calculate the "dissimilarity matrix" (or simply: D matrix) based on the navigator data, which is additionally collected in addition to the normal imaging data.

2. Characterize motion by quantitative analysis of the D matrix. This is possible because different types of motion lead to eigenmodes in the D matrix. Furthermore, the strength of the pattern reflects the strength of the movement, so that quantitative conclusions can be drawn from the evaluation of the matrix.

To implement an embodiment, navigator data may be collected in addition to normal imaging data. The essential characteristic that the signal must have to qualify as a navigator is that it should be reproducible in the absence of motion.

Many options exist to achieve this, such as:

- Adds an echo to TSE or TFE acquisitions to acquire Classic Navigators, or Floating Navigators, or Orbiting Navigators.

- Use FID as navigator

- Navigators can also be inserted into FFE sequences.

- Forming the navigator as an element in an appropriate magnetization preparation scheme either before or after the data acquisition sequence.

The computation of the dissimilarity matrix is discussed next. One element D _ij of the dissimilarity matrix is computed from navigator data by comparing navigator data i with navigator data j (eg by computing the sum of the squared composite differences over all navigator samples). This example is exemplary only, and there are alternative ways of computing D _ij . If N navigators were collected during data collection, this results in an NxN matrix.

Once the dissimilarity matrix has been calculated, different motion types produce different patterns on the D matrix. This will be shown below with a few examples from volunteer experiments. Analyzing these patterns allows deciding whether reacquisition of certain data prevents image artifacts due to motion.

4 to 11 , 14 and 15 illustrate visualizations of similarity matrices. In these examples, the fading of the squares is used to indicate dissimilarity. If the square is completely white, it has low dissimilarity. If the square is darker, it has high dissimilarity. All examples are from multi-slice, multi-shot TSE sequences where orbital navigators are added to the echo chain. The D-matrix is shown as a gray scale, where dark values represent large dissimilarities and white represent low dissimilarities.

First, the detection of reversible accidental motion is illustrated. FIG. 4 shows a visualization of a dissimilarity matrix 400 for an example from T2-weighted head imaging. The x-axis is labeled 402 and the y-axis is labeled 404 . Both the x-axis and the y-axis indicate the number of shots. Excitations as used herein are understood to be equivalent to portions of magnetic resonance data. Most of the D matrix is white, which indicates low dissimilarity. Most of the Navigator and most of the TSE excitations agree well with each other. The exception is shot number 5 (406) which differs from all other shots. In this experiment, volunteers briefly scratched their foreheads during data acquisition, causing the challenge number 5 (406) to be disturbed. Image artifacts result from this interference. If the shot number 5 (406) is replaced with reacquired data, the artifact disappears. Other examples for this type of movement are coughing and swallowing.

Next, the detection of periodic motion is illustrated. FIG. 5 shows another visualization of a dissimilarity matrix 500 . This example is also from T2-weighted head imaging. Here, the D matrix can be divided into two groups, where the excitations within one group are in good agreement with the excitations of the other group, but not with the excitations of the other groups: Group 1 = 1, 5, 9, 14, 18, 19 ; Group 2 = 2, 3, 4, 6, 7, 8, 10, 11, 12, 13, 15, 16, 17. This results in a checkerboard-like pattern. In this example, the motion is the pulsation of blood vessels. The regular pattern is caused by beating at the heart rate and the navigator rate (determined by sequence TR). Reacquisition of the excitations in group 1 can reduce pulse artifacts in the images. Similar patterns may result from different appropriate superpositions of part-period motion sources such as respiration and cardiac motion.

FIG. 6 shows another visualization of a dissimilarity matrix 600 constructed during T1-weighted shoulder imaging and also illustrating periodic motion. It shows a fairly regular checkerboard pattern induced by breathing motion. Superimposed on the checkerboard is a progressively increasing dissimilarity (= progressively increasing time between navigators) towards the upper right and lower left corners of the D-matrix. This is caused by drift.

This is also the first example of the ability of the D-matrix to be used to distinguish between different types of motion, in this case drift and periodic motion. Based on these D-matrix data, a decision can be made whether re-acquisition is meaningful to potentially improve data consistency and image quality.

Next, the detection of sudden irreversible changes is illustrated. Fig. 7 shows another example of a dissimilarity matrix acquired during T1-weighted shoulder imaging. The abrupt change in the dissimilarity matrix is shown around firing number 95 (labeled 702), where a sudden irreversible change occurs. All the data from the start of the scan up to that time agrees well with the other, but doesn't match anything that happened after that event. All data measured after challenge #95 were in good agreement with each other and did not match the first part of the scan. This is the case where reacquiring several shots will not prevent artifacts. Instead, all firing numbers 1 to 95 are preferably repeated.

It should be noted that the D matrix also weakly contains a checkerboard-like breathing motion pattern. But its intensity is relatively small compared with sudden irreversible movement.

Next, the detection of combinations of reversible and irreversible motions is illustrated. FIG. 8 shows another visualization of a dissimilarity matrix 800 . In this example (T1w shoulder imaging) a sudden irreversible motion is shown at shot number 130, 802, and a short period of reversible motion around number 170-180 (labeled 804). In addition, there is a weak regular checkerboard pattern due to the breath present for the entire scan.

FIG. 9 shows another visualization of a dissimilarity matrix 900 illustrating how image quality can be improved by analyzing the dissimilarity matrix 900 . Figure 9 is another example why the D-matrix can be useful, the figure below shows the results from head imaging where a volunteer coughed gently during the scan. The left image shows the D matrix being analyzed in real time to reacquire the disturbed data (in this example shot number 1 - shot number 4, which is labeled 902). The center and right images show reconstructed images without and with reacquired data, respectively. Recapturing significantly improves image quality by reducing ghosting artifacts.

The following figures illustrate the application of cluster analysis to embodiments of the invention. 10 and 11 show visualizations of the D-matrix 1000 , 1100 .

Figure 10 shows an example of a D-matrix 1000 for an experiment consisting of 19 TSE excitations. The fluctuations in the dissimilarity value are not random, but appear to follow a checkerboard-like pattern. This becomes apparent when the TSE excitations are grouped into clusters and the rows and columns of the D-matrix are reordered according to the order created by the clustering process.

A reordered D-matrix 1100 is shown in FIG. 11 . From the reordered matrix it is evident that the data set can be divided into two different groups.

In the example, the "Agglomerative Hierarchical Clustering Algorithm with Average Linkage" is used. This is an algorithm that creates a hierarchical sequence of clusters from a finite set of n elements: the algorithm is initialized by creating n clusters each containing only 1 element. Then at each step, the two clusters with the least dissimilarity are merged. The value of the dissimilarity at which two clusters are merged is stored by the algorithm (hereinafter referred to as the fusion threshold). Since the cluster with the lowest dissimilarity is always fused, the fusion threshold decreases monotonically with the number of clusters. After n steps, the algorithm ends since all clusters have been merged into one cluster containing the entire set.

A more detailed analysis of the clustering process can be used to characterize the dataset. This will be discussed below using a few examples. The fusion threshold is a measure of the average dissimilarity for all elements within a cluster that are considered still admissible. The thresholds to be rated as permissible can be determined either by domain-specific knowledge (ie the observed distribution of D-matrix values from previous experiments), or a model of the underlying process (MR signal strength, noise, . . . ). Yet another alternative, independent of the absolute threshold, is to consider a fusion threshold that is related to the number of clusters. The optimal number of predicted clusters is indicated by the 'kinks' in the plot.

FIG. 12 shows a graph of the number of clusters 1200 versus fusion threshold 1202 for the examples shown in FIGS. 10 and 11 . It is evident from this that there is a large drop (410 → 280) in the fusion threshold when going from one to two clusters. From two to three, the threshold drops from 280 to 220. Thereafter, there is a gradual drop in threshold. Arguably the optimal number of clusters is 2 or 3. More insight on how to solve this problem can be learned from different visualizations of the clustering process, which contain more information:

FIG. 13 plots cluster size distributions as a function of dissimilarity level 1300 . Figure 13 shows which distribution of cluster sizes exists at different levels of inter-cluster dissimilarity. On the horizontal axis 1300 the value of the dissimilarity is shown. One unit on vertical axis 1302 corresponds to one element in the set. Elements belonging to one cluster are shown in the same gray scale. The distribution of cluster sizes at a certain level of dissimilarity can be found by drawing vertical lines at the respective levels. The number of clusters present is given by the number of different gray levels below the line. The size of each cluster is the length of intersection of the line with the gray scale.

For example, at a dissimilarity level of 400, there are only two clusters. The first cluster 1304 contains 9 elements and the second cluster 1306 contains 10 elements.

If the dissimilarity threshold is lowered, a first change occurs at the level of 280: here the second cluster is split into two subgroups, one containing 2 and the other 8 elements.

In this context, the aim of cluster analysis is to improve the quality of MR datasets by reacquiring data disturbed by motion. That is, a trade-off must be made between additional imaging time and possible improvement in quality.

From Fig. 13, it can be seen that if all elements of the first cluster are recollected (and the recollected data belong to the second cluster), the dissimilarity of the entire dataset can be reduced from 400 to 280. A further reduction in dissimilarity is possible by resampling 2 elements of a smaller subgroup of the second cluster. This means that the level of clustering can be used to map how much improvement can be expected with how much extra scan time. That is, the real question is not where to put the threshold for clustering, but how much extra time can be spent on some quality improvement. This is a decision about the trade-off between speed and quality that will be made by the operator of the MR system via the scanner's user interface.

Predictions of possible improvements cannot be made from Figure 13 alone, because an important piece of information is missing in this visualization: the temporal ordering of the elements.

14 and 15 show further visualizations of D matrices 1400 , 1500 . These figures show the observed D-matrix 1400 and the result after reordering 1500 (compared to the previous example, in which there are more rows and columns because a different sequence was used in this experiment, which was into more TSE segments).

In this example, the data set can also be split into two groups of roughly equal size. But in this paradigm, the elements belonging to the two clusters are continuous in time, while in the first paradigm, they are not. Fig. 16 shows the cluster index for different TSE segments of the D matrix shown in Fig. 14, where the index oscillates between two clusters. This pattern is characteristic of periodic motion between two states.

FIG. 17 shows cluster indices for different TSE segments of the D matrix shown in FIG. 15 . The patterns in Figure 17 are very different from those in Figure 15: the first part of the scan belongs to the first cluster, the second part to the other. This is characteristic of irreversible changes during scanning. If it were decided to eliminate elements of cluster 2 by reacquisition, there would be no hope of success as in this paradigm the patient would be very unlikely to return to the first motion state.

This qualitative discussion can be made more quantitative by calculating the probability that the state of motion will change from one state to another.

Figure 18 shows the evolution of the transition probability distribution (PDF) during the experiment. The left image 1800 shows the probability of transition from state 1 to state 2 and the right image 1802 shows the probability of transition from state 2 to state 1 . Each horizontal line in the image shows a probability density function for a state transition, coded in grayscale (white represents 0). As these probability distributions evolve each time the patient is in a certain motion state at some time point i, the observation of whether he is still in the same motion state at time point i+1 can be used to update the probability distribution function. That is, the number of lines in each image is given by the number of times the patient was observed in the respective motion state.

This means that the last line in each image represents the state with knowledge about the probability of the state transition at the end of the experiment.

Figure 19 shows on the left the situation for this pair of first paradigms, where repeated state transitions occur during the experiment. Therefore, the probability for an additional state transition is estimated to be approximately 50%. That is, an attempt to use reacquisition to replace data from cluster 1 with data from cluster 2 in this example was successful.

In contrast, Fig. 19 shows on the right the same situation for the second example, where only one state transition occurs at -60% of the measurement time. Both PDFs here at the end of the experiment show that state transitions are highly improbable. That is, attempting to replace data from cluster 2 by reacquisition will not be successful. In this paradigm, different strategies are required: for example, readjust the scan geometry to cover the position of the patient in state 1, or abort the scan, or notify the user. This is just one example of how clustering can help in making decisions to address motion-related quality issues.

FIG. 19 illustrates the evolution of indications for transition probabilities between two observed motion states in the two examples shown in FIG. 18 . The bottom line in each image represents the state of knowledge at the end of the experiment.

Left panel 1900: Data for the first paradigm. Here, the PDF is quite broad and centered at 50% of the end of the experiment, indicating that there is a fair chance that the patient will change between exercise states in the future.

Right panel 1902: Data for the second paradigm. Both PDFs quickly evolved into narrow distributions at low probability values due to the fact that one state transition was detected during the entire experiment.

While the invention has been illustrated and described in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive; the invention is not limited to the disclosed embodiments .

Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. The computer program may be stored/distributed on suitable media, such as optical storage media or solid-state media provided with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication system. Any reference signs in the claims should not be construed as limiting the scope.

Claims (16)

1. A magnetic resonance imaging system (200, 300) for acquiring magnetic resonance data (242, 244) from an imaging region (208), wherein the magnetic resonance imaging system comprises: - a processor (230) for controlling the magnetic resonance imaging system; and - memory (236) for storing machine-executable instructions (250, 252, 254, 256, 258) for execution by said processor, wherein execution of said machine-executable instructions causes said processor repeatedly: - controlling (100) said magnetic resonance imaging system to acquire portions of said magnetic resonance data, wherein each portion of said magnetic resonance data comprises navigator data (244); - creating (102) a set of navigator vectors by extracting said navigator data from each portion of said magnetic resonance data; - constructing (104) a dissimilarity matrix (246, 400, 700, 800, 900, 1000, 1100, 1400, 1500) by computing a measure between each of said navigator vectors in said set of navigator vectors ; - using a classification algorithm to generate (106) a matrix classification (248) of said dissimilarity matrix; and - controlling (108) said magnetic resonance imaging system to modify acquisition of said magnetic resonance data using said matrix classification. 2. The magnetic resonance imaging system of claim 1, wherein the classification algorithm is a pattern recognition algorithm operable to select the matrix classification. 3. The magnetic resonance imaging system of claim 2, wherein the memory further comprises a matrix library (310) comprising example matrices, wherein the pattern recognition algorithm can be used to select the One of the example matrices described above. 4. The magnetic resonance imaging system of claim 3, wherein each of said example matrices is associated with a modification instruction (312), wherein a modification of said magnetic resonance data by said magnetic resonance imaging system Acquisitions are modified by executing said modification instructions. 5. The magnetic resonance imaging system of claim 2, wherein the pattern recognition algorithm is a cluster analysis algorithm, wherein the cluster analysis algorithm is operable to perform a temporal correlation of the set of navigator vectors. 6. The magnetic resonance imaging system of claim 1, wherein the classification algorithm is a statistical analysis algorithm. 7. The magnetic resonance imaging system of claim 6, wherein the statistical analysis algorithm is capable of determining the matrix classification by performing any of the following: performing Bayesian analysis; analyzing the dissimilarity thresholding the matrix; calculating a standard deviation of the dissimilarity matrix; identifying elements in the dissimilarity matrix that are outside a predetermined range; and performing probability-based selection. 8. The magnetic resonance imaging system of any one of the preceding claims, wherein the magnetic resonance data comprises a plurality of slices, wherein execution of the instructions further causes the processor to: - computing a set of dissimilarity matrices using said dissimilarity matrix for each of said plurality of slices; - generating a set of matrix classifications by using said classification algorithm to generate said dissimilarity matrices for each of said set of dissimilarity matrices; and - controlling the magnetic resonance imaging system to modify the acquisition of the magnetic resonance data using the set of matrix classifications. 9. The magnetic resonance imaging system of claim 1, wherein the magnetic resonance imaging system further comprises a multi-channel radio frequency system operable to receive the magnetic resonance data from more than one channel simultaneously , wherein execution of the instructions further causes the processor to create the set of navigator vectors by combining the navigator data from the more than one channel. 10. The magnetic resonance imaging system of claim 9 , wherein the navigator data is combined by using any of the following: pairing the navigator data from the more than one channel with a predetermined weighting averaging and concatenating said navigator data from said more than one channel. 11. The magnetic resonance imaging system of claim 1 , wherein the acquisition of magnetic resonance data is modified by any of the following manners: stopping the acquisition of magnetic resonance data; modifying scan geometry and restarting the acquisition of magnetic resonance data; ignoring one or more portions of the magnetic resonance data; reacquiring the portion of the magnetic resonance data; generating an operator alert; and combinations thereof. 12. The magnetic resonance imaging system of claim 1, wherein the metric is any one of: computing a sum of squared composite differences between navigator vectors; computing a difference in magnitude of the navigator vectors ; calculate the absolute value of the difference between the navigator vectors; and, calculate the correlation between the navigator signals. 13. The magnetic resonance imaging system of claim 1 , wherein the magnetic resonance data comprises image data, wherein execution of the instructions causes the magnetic resonance imaging system to acquire the image data from a first region of interest And collecting said navigator data from a second region of interest. 14. The magnetic resonance imaging system of claim 1, wherein the magnetic resonance imaging system includes a motion detection system (302) for acquiring motion data (314), wherein execution of the instructions causes the processing acquires the motion data during the acquisition of the magnetic resonance data, wherein execution of the instructions further causes the processor to incorporate the motion data into the dissimilarity matrix. 15. A magnetic resonance imaging method for acquiring magnetic resonance data (242, 244) from an imaging region (208), wherein the magnetic resonance imaging method comprises repeatedly performing the steps of: - controlling (100) a magnetic resonance imaging system to acquire portions of said magnetic resonance data, wherein each portion of said magnetic resonance data comprises navigator data; - creating (102) a set of navigator vectors by extracting said navigator data from each portion of said magnetic resonance data; - constructing (104) a dissimilarity matrix (246, 400, 700, 800, 900, 1000, 1100, 1400, 1500) by computing a measure between each of said navigator vectors in said set of navigator vectors ; - using a classification algorithm to generate (106) a matrix classification of said dissimilarity matrix; and - controlling (108) said magnetic resonance imaging system to modify acquisition of said magnetic resonance data using said matrix classification. 16. A computer readable medium having stored thereon machine readable instructions (250, 252, 254, 256, 258), the machine readable instructions comprising: - machine readable instructions for controlling (100) a magnetic resonance imaging system to acquire portions of magnetic resonance data, wherein each portion of the magnetic resonance data includes navigator data; - machine readable instructions for creating (102) a set of navigator vectors by extracting said navigator data from each portion of said magnetic resonance data; - for constructing (104) a dissimilarity matrix (246, 400, 700, 800, 900, 1000, 1100, 1400, 1500) machine-readable instructions; - machine readable instructions for generating (106) a matrix classification of said dissimilarity matrix using a classification algorithm; and - Machine readable instructions for controlling (108) said magnetic resonance imaging system to modify acquisition of said magnetic resonance data using said matrix classification.